A Predictive Model for Cancer and Non- Cancer Mortality in Beagle Dog Study
View Presentation View Presentation
*Changsheng Lao, FDA\CDRH 


The Weibull and Gompertz probability distributions were applied to fit the 1680 beagle dogs received radiation (sham, low and high doses) for mortality from cancer and non-cancer deaths. The Cox proportional hazard regression model was used to identify prognostic factors to predict future dog deaths. The fitted parametric models were used to project future dog deaths from cancer and non-cancer mortality for the next five years in order to allocate budget of research for this project. Our statistical analyses applied five important procedures to project future mortality: Model identification, model fitting, diagnostic check, forecasting and validation.